- Email: [email protected]
Client characteristics, organizational variables and burnout in care staff: The mediating role of fear of assault
Research in Developmental Disabilities 34 (2013) 940–947
Contents lists available at SciVerse ScienceDirect
Research in Developmental Disabilities
Client characteristics, organizational variables and burnout in care staff: The mediating role of fear of assault John Rose *, Sophie Mills 1, Daniel Silva 2, Lauren Thompson 3 The University of Birmingham, Edgbaston, Birmingham, B15 2TT, United Kingdom
A R T I C L E I N F O
A B S T R A C T
Article history: Received 24 September 2012 Received in revised form 19 November 2012 Accepted 19 November 2012 Available online 3 January 2013
A broad range of factors have been identified as having an impact on burnout and performance. To improve our understanding of how these factors interact, a model of carer stress is tested. Staff were surveyed in residential units, assessments included burnout, organizational factors, staff cognitions and ratings of resident challenging behavior. The relationship between challenging behavior and emotional exhaustion was fully mediated by fear of assault. The relationship between emotional exhaustion and experienced safety (an organizational variable) was also fully mediated by fear of assault. The use of the model with staff is supported and it suggests that staff burnout can be reduced by influencing either staff cognitions, organizational factors or challenging behavior or a combination of these factors. ß 2012 Elsevier Ltd. All rights reserved.
Keywords: Burnout Challenging behavior Staff Cognitions Organizational factors Intellectual disability
1. Introduction Staff are increasingly being recognized as an important element in the care process (National Health Service, 2009). The psychological well-being of staff is seen as an important element in the delivery of care and support, and there is increasing evidence of a link between staff well-being and their effectiveness in providing support to people with intellectual disabilities (e.g. Hatton, Wigham, & Craig, 2009; Rose, Jones, & Fletcher, 1998). A range of variables have been identified as contributing to staff well being including client characteristics (e.g. Jenkins, Rose, & Lovell, 1998), organizational factors and staff cognitive variables (e.g. Rose, 2011). There is evidence to suggest that the way that staff perceive aspects of their working life, particularly challenging behavior can influence their reported levels of burnout. Howard, Rose, & Levinson (2009) have shown the relationship between challenging behavior and burnout can be moderated by the perceived self efficacy of staff. In a study based on a subset of the data presented here Mills and Rose (2011) have also shown the same relationship can be mediated by perceived fear of assault. That is staff report lower levels of burnout if they feel more competent, supported and consider the possibility of them being assaulted as lower than other staff. A similar relationship was found by Rose, David, and Jones (2003) in that wishful thinking (a coping strategy) partially mediated the relationship between work demands (of which challenging behavior could be an integral part) and stress. This relationship was confirmed by Devereux, Noone, Firth, and Totsika (2009) who found that wishful thinking was partially mediated by the relationship between work demands & emotional exhaustion in care staff.
* Corresponding author at: School of Psychology, The University of Birmingham, Edgbaston, Birmingham, B15 2TT, United Kingdom. Tel.: +44 01214142640. E-mail address: [email protected] (J. Rose). 1 Now at Black Country Partnership Foundation (NHS) Trust. 2 Now at University of Southampton. 3 Now at Bolton NHS foundation Trust. 0891-4222/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ridd.2012.11.014
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
941
Other studies have suggested that organizational variables contribute significantly to stress and burnout (Hatton, Brown, Caine, & Emerson, 1995; Thompson & Rose, 2011), and recent research that evaluates the differences between a variety of inputs and outputs in relation to psychological well-being suggests that staff rank organizational factors as the single largest contribution to staff well-being (Rose, Madurai, Thomas, Duffy, & Oyebode, 2010; Thomas & Rose, 2010). A complex array of factors can contribute to reported well-being and this complexity has contributed to the lack of a comprehensive model to explain the development of staff stress. A number of frameworks and models have been developed including Rose (1995) who proposed a simple framework relating to organizational proximity, and Hatton, Rose, and Rose (2004) who listed a broad range of core constructs that might influence the development of well-being, suggesting that staff morale and well-being was one construct that, among many, may directly influence well being. Other models have examined specific aspects of the relationship between staff and the people they care for such as attributions and have used Weiner’s (1985) model however, the results from these studies have been mixed (Dagnan, Trower, & Smith, 1998; Rose & Rose, 2005). Other models have been developed for specific purposes such as examining the importance of staff characteristics (Rose et al., 2003) or identifying factors that may be responsible for placement breakdown (Phillips & Rose, 2010). These models tend to be specific in focus and lack a broader overview so while they can contribute to answering specific questions they lack a broader conceptual overview. Alternative models can be found that examine the development of stress in parents of children with intellectual disabilities (e.g. Hill & Rose, 2010). While some of these models are complex and have been difficult to test empirically others have been partially tested and the results suggest that they have some utility. For example, a model of stress in parent child interactions based on Mash and Johnston (1990) is shown in Fig. 1. Mash and Johnston’s (1990) model consists of four elements including child characteristics such as ability and challenging behavior, environmental characteristics such as family support, parent characteristics that might include a variety of psychological variables such as attributions or locus of control with parent child interactive stress as an outcome. They see these links as reciprocal and while they recognize that other variables may be influential they note that these variables are likely to be the most important influence on the development of parent–child interactive stress. This model has been demonstrated to have predictive utility in mothers of children with intellectual disability (Hassall, Rose, & McDonald, 2005; Hill & Rose, 2009). A review of this model suggests that it may have direct applicability to understanding the development of staff well-being and burnout in that elements of the model map closely on to the factors that have been identified as important in research on staff well-being. The characteristics of clients are clearly of direct relevance, particularly challenging behavior.
Fig. 1. A model of stress in parent child interactions based on Mash and Johnston (1990).
942
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
Environmental or organizational characteristics have also been identified as important variables for staff (Thompson & Rose, 2011). A variety of staff variables have also been identified as important mediators between challenging behavior and wellbeing such as attributions, self efficacy and fear of assault. If effective measures of these elements of the model can be taken then their relationships can be investigated in relation to a modified version of the Mash and Johnston (1990) model. There are a range of commonly used measures of challenging behavior for people with Intellectual Disability (e.g. Checklist of Challenging Behaviors (Harris, Humphreys, & Thomson, 1994). A number of staff cognitive variables have also been shown to mediate the relationship between stress and challenging behavior including fear of assault, this is a concept developed by Leather, Beale, Lawrence, and Dickson (1997) who investigated levels of violence and fear of violence experienced by publicans in relation to their well-being. Measures of organizational characteristics are also being developed including the Essen Climate Evaluation Schema (EssenCES) which is a measure designed to assess organizational characteristics such as the social climate in forensic settings (Schalast, Redies, Collins, Stacey, & Howells, 2008) this assesses dimensions of social climate, including three subscales: ‘therapeutic hold’, ‘patient cohesion and mutual support’ and ‘experienced safety’. While these variables may not comprehensively represent all of the concepts originally considered by Mash and Johnston, they do provide a starting point for the empirical testing of the model. 1.1. Aim The aim of this paper is to use the structure provided by the Mash and Johnston (1990) model with direct care staff in residential settings. The model will be used to examine the relationship between staff psychological well being, challenging behavior and the therapeutic environment and how the relationship between these variables is mediated by a psychological variable, fear of assault. By doing this it is hoped that it will provide an initial test of the Mash and Johnston (1990) model with a view to gaining a greater understanding of how these complex factors interact. 2. Method 2.1. Design and sample The study was reviewed and given a favorable ethical opinion by a National Health Service (NHS) Local Research Ethics Committee. A cross-sectional correlational design was used. The participants included qualified and unqualified direct care staff who worked in residential homes for adults with intellectual disabilities. All of the staff were involved directly in the care of people with intellectual disabilities and would have spent the majority of their work time working with clients. Staff were required to have worked in their current place of work for at least three months. Staff working in a variety of services including both the National Health Service (NHS) and private sector were recruited, 78 staff took part from a total of six organizations however one was removed from the analysis as they had been at work for less than three months. The majority of staff worked in community settings, 14% of staff worked in secure settings. 2.2. Procedure Clinical psychologists were asked to identify homes where service users displayed challenging behavior and managers were approached to invite staff to take part in the study. While the focus of the study was on challenging behavior, in practice there was a very broad range of environments sampled from locations where considerable challenging behavior was experienced to other environments where residents rarely exhibited any challenging behavior. Questionnaire packs were then left with managers to distribute to the staff. Packs included a Participant Information Sheet, the Consent Form, and questionnaire. The questionnaire pack took approximately 30 min to complete. Participants either posted the questionnaires back to the researchers using a pre-paid envelope, or they were collected from the homes. 333 questionnaires were distributed and 78 were returned, a response rate of 23%. 2.3. Measures Fear of assault. The fear of assault measure comprised of two questions to measure staff’s fear of assault rated on a 5-point likert scale, adapted from Leather et al. (1997) by Rose and Cleary (2007). A Checklist of Challenging Behaviors. The Checklist of Challenging Behaviors (Harris et al., 1994), measures challenging behavior exhibited by a service user in the last three months. The participant is asked to rate on 5 point scales the frequency, management difficulty, and severity of 14 ‘aggressive behaviors’ (e.g. scratching, biting, kicking people) and 18 ‘other challenging behaviors’ (e.g. smashing windows, spitting, eating inappropriate things). For these other challenging behaviors the participant is only asked to rate the frequency and management difficulty, not severity. The checklist is a reliable measure, with significant critical values for rs at p < 0.05 using Spearman’s correlation coefficient for frequency, management difficulty, and severity. Maslach Burnout Inventory – Human Services Survey (MBI-HSS). (Maslach, Jackson, & Leiter, 1996) measures burnout in staff. Respondents complete statements rated from 0 (never) to 6 (every day) regarding their feelings about their
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
943
employment. The measure is divided into three scales; emotional exhaustion (EE), depersonalization (DP), and personal accomplishment (PA). A high level of burnout is characterized by high EE and DP scores, and low PA scores. Hastings, Horne, and Mitchell (2004) carried out a factor analytic study with staff working with people with intellectual disabilities the integrity of the scale and reliability were found to be acceptable (alpha: EE = 0.87, DP = 0.68, PA = 0.76). Modified version of Essen Climate Evaluation Schema. (EssenCES; Schalast et al., 2008). The EssenCES is a 15-item questionnaire that also uses a 5 point scale and provides a score related to the social climate in 3 domains: ‘therapeutic hold’, ‘patients’ cohesion and mutual support’, and ‘experienced safety.’ High scores in each of the domains represent a positive social climate. The EssenCES was designed to assess the social climate of forensic psychiatric wards, but it has been suggested that it could also be used in similar settings, including environments for people with challenging behavior. The questionnaire was adapted for example; the word ‘‘resident’’ was used instead of ‘‘patient’’. The unmodified EssenCES has reasonable internal consistency (therapeutic hold, a = 0.74; patient cohesion and mutual support, a = 0.78; experienced safety, a = 0.77; Howells et al., 2009). 2.4. Reliability To assure the internal consistency of the questionnaires and the ecological validity of this analysis, reliability was checked for two measures: EssenCES and the fear of assault. The Cronbach’s alpha for the three components of the EssenCES questionnaires were: therapeutic hold (0.73), experienced safety (0.80) and patient cohesion (0.80). In order to account for the number of items in the fear of assault questionnaire an inter item reliability was calculated that showed a positive correlation of r = 0.89. 3. Results The final sample consisted of 77 participants. There were also some missing data. See Tables 1 and 2 for demographic data from the sample. Means and standard deviations are reported for the main study variables in Table 3. Shapiro–Wilk tests were conducted and two of the scales failed the normality assumption: Patient Cohesion from EssenCES and other challenging behavior from the Checklist of Challenging Behavior. Table 1 Staff demographics: age and length of time in current and previous employment in intellectual disability services.
Age (years) Length of time in current employment (months) Length of time working in services for people with intellectual disabilities (months)
Mean
Range
SD
N
37 62 101
18–62 3–279 3–387
11.89 75.32 97.89
76 76 76
Table 2 Demographic data: gender, job title and training.
Gender Job title
Training related to challenging behavior
Male Female Support workers Qualified nursing staff Managerial positions (e.g. house/team leader) Other disciplines (including psychology, occupational therapy, social work) Breakaway/de-escalation Training relevant to challenging behavior (unspecified)
Percentage
N
29.87 70.13 63.16 11.84 14.47 10.53 19.05 4.76
77 76
63
Table 3 Means and standard deviations of scales.
Fear of assault Emotional exhaustion Depersonalisation Personal accomplishment Aggressive behavior Other behavior Patient cohesion Experienced safety Therapeutic hold
Minimum
Maximum
Mean
Std. deviation
2.00 0.00 0.00 19.00 1.00 1.00 0.00 0.00 7.00
10.00 51.00 26.00 48.00 3.98 4.22 19.00 18.00 20.00
5.3117 19.8649 4.7203 36.3959 2.1139 2.1611 8.6554 8.0682 15.7829
2.01481 11.09155 5.37931 6.67392 0.80512 0.70115 4.58028 4.75028 3.00343
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
944 Table 4 Correlation matrix for study variables. (1)
Measure Fear of assault (1) Emotional exhaustion (2) Depersonalization (3) Personal accomplishment (4) Aggressive behavior (5) Other behavior (6) Patient cohesion (7) Experienced safety (8) Therapeutic hold (9)
0.392
(2) **
(3) **
0.465 0.337**
(4) **
0.275 0.323** 0.206
(5) **
0.560 0.364** 0.295** 0.384**
(6) **
0.600 0.329** 0.410** 0.184 0.688**
(7) *
0.244 0.339** 0.219 0.343** 0.350** 0.369**
(8) **
0.451 0.266* 0.218 0.213 0.412** 0.439** 0.013
0.187 0.095 0.185 0.352** 0.268* 0.247* 0.378** 0.043
* Significant at P < 0.05. ** Significant at P < 0.01.
The analysis was performed in three steps. The first step considered the univariate relationships between the variables. Where the variables were normally distributed Pearson’s correlation was used and where the variables were not normally distributed Spearman’s Rho was calculated (Table 4). Second, an attempt was made to construct a pathway linking burnout, challenging behavior, environment measures and cognitive variable. Third, where the relationships indicated possible mediation, these processes were investigated. 3.1. Path analysis In order to construct a model of burnout considering all of the variables, the correlations between the measures for each variable were analyzed. The goal was to find pathways of significant correlations that linked measures of challenging behavior and environmental variables with the cognitive variable and burnout and that also linked the cognitive variable to burnout (see Fig. 2). A model would be formed if one of the measures from each of the variables followed these specifications. The data produced five models that fulfilled these conditions. The models are shown on Fig. 2 below.
Fig. 2. Adapted model of stress for staff–resident interactions showing only significant paths.
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
945
Table 5 Components of the models formed with the data by measure. Model
Challenging behavior
Organizational characteristics
Cognitive variables
Burnout
1 2 3 4 5
Aggressive Aggressive Aggressive Other Other
Patient cohesion Patient cohesion Experienced safety Patient cohesion Experienced safety
Fear Fear Fear Fear Fear
Emotional exhaustion Personal accomplishment Emotional exhaustion Emotional exhaustion Emotional exhaustion
of of of of of
assault assault assault assault assault
3.2. Mediation Although the models indicate a relationship between the variables, a formal analysis for mediation is necessary to accurately understand their scope. Mediation analyses are often guided by the assumptions proposed by Baron and Kenny (1986): the independent variable (IV) significantly affects the mediator, the IV significantly affects the dependent variable (DV) in the absence of the mediator, the mediator has a significant unique effect on the DV and finally, the effect of the IV on the DV reduces upon the addition of the mediator to the model. However, Baron & Kenny’s model for mediation is quite conservative in defining full mediation as when the effect of the IV on the DV in the presence of the mediator diminishes to zero. This scenario is unrealistic for the social sciences (Jose, 2008). Thus, it is recommended that mediational analyses be based on formal significance tests of the indirect effect. The most used formal analysis of mediation is the Sobel test (Sobel, 1982), this approach is more powerful than the stepwise procedure proposed by Baron and Kenny because it addresses mediation directly. Although useful, for this particular analysis the Sobel test cannot be used because the variables are not normally distributed. Therefore, bootstrapping is recommended. In such procedure, random amounts of indirect effects are extracted from the data and the confidence interval for each of these samples is calculated to inform what would be the confidence interval for the whole data. Through the application of bootstrapped confidence intervals, it is possible to avoid power problems introduced by asymmetric sampling distributions of an indirect effect (MacKinnon, Lockwood, & Williams, 2004). Preacher and Hayes (2004) developed a macro script to formally evaluate full mediation through bootstrapping. The matrix gives you the result for the four assumptions from Baron and Kenny as well as the confidence interval for both 95% and 99% confidence. With such analysis to reject the null hypothesis (assume significant mediation) zero cannot be within your confidence interval. This would mean that there is a possibility that the mediation effect could be due to chance. In order for the macro script to be used, the models were split in two subsets. On one side, the direct and indirect effect of challenging behavior on burnout and on the other side the direct and indirect effect of environment on burnout. This was Table 6 Summary of regression analysis between experienced safety and emotional exhaustion mediated by fear of assault. Regression
Independent variables
Dependent variables
1 2 3
Experienced safety Fear of assault Experienced safety Fear of assault Emotional exhaustion experienced safety
Emotional exhaustion Emotional exhaustion Emotional exhaustion
0.12 (0.05) 0.08 (0.02) 0.99 (0.30)
2.61 4.33 3.40
0.01** 0.00** 0.00**
Fear of assault
0.049 (0.05)
0.94
0.35
4 Bootstrapping effect
Mean 0.072
SE 0.028
Beta (S.E.)
t
p
LL 99% CI
UL 99% CI
0.166
.013
Note. n = 77 participants. Unstandardized regression coefficients are reported. Bootstrap sample size = 10,000. LL, lower limit; CI, confidence interval; UL, upper limit.
Table 7 Summary of regression analysis between aggressive behavior and emotional exhaustion mediated by fear of assault. Regression
Independent variables
Dependent variables
Beta (S.E.)
t
p
1 2 3
Aggressive behavior Fear of assault Aggressive behavior Fear of assault Emotional exhaustion aggressive behavior
Emotional exhaustion Emotional exhaustion Emotional exhaustion
0.03 (.07) 0.08 (.02) 0.18 (.04)
4.09 4.33 4.23
0.00** 0.00** 0.00**
Fear of assault
0.02 (.00)
2.11
0.03**
4
Bootstrapping effect
M
SE
LL 99% CI
UL 99% CI
0.154
0.005
0.003
0.032
Note. n = 77 participants. Unstandardized regression coefficients are reported. Bootstrap sample size = 10,000. LL, lower limit; CI, confidence interval; UL, upper limit.
946
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
Table 8 Summary of regression analysis between other challenging behavior and emotional exhaustion mediated by fear of assault. Regression
Independent variables
Dependent variables
Beta (S.E.)
t
p
1 2 3
Other challenging behavior Fear of assault Other challenging behavior Fear of assault Emotional exhaustion other challenging behavior
Emotional exhaustion Emotional exhaustion Emotional exhaustion
0.02 (0.01) 0.08 (0.02) 0.19 (0.03)
3.55 4.33 5.29
0.00** 0.00** 0.00**
Fear of assault
0.01 (0.01)
1.23
0.22
4 Bootstrapping effect
M
SE
LL 99% CI
UL 99% CI
0.160
0.005
0.005
0.031
Note. n = 77 participants. Unstandardized regression coefficients are reported. Bootstrap sample size = 10,000. LL, lower limit; CI, confidence interval; UL, upper limit.
necessary because the program could only account for one dependent variable, one independent variable and one mediator at a time. However, for the interpretation of the results, a model would only be considered significant if both subsets showed significant results. The bootstrapping results showed no significant mediation between the following subsets: patient cohesion – fear of assault – emotional exhaustion (CI at 95% = 0.07, 0.02) and patient cohesion – fear of assault – personal accomplishment (CI at 95% = 0.01, 0.07). Due to these results, models one, two and four (Table 5) were disregarded. Tables 6–8 show the results for the three regressions suggested by Baron and Kenny and the bootstrapping analysis for the significant subsets. The significant indirect effect indicates full mediation for the following models: aggressive challenging behavior, experienced safety, fear of assault and emotional exhaustion; and other challenging behavior, experienced safety, fear of assault and emotional exhaustion. 4. Discussion The results suggest that Mash and Johnston’s (1990) model may have utility as burnout can be understood as a corollary of challenging behavior, organizational climate and cognitive variables. The influence of challenging behavior including both aggression and other challenging behavior, as measured by the Checklist of Challenging Behavior on emotional exhaustion is fully mediated by fear of assault. However, only the relationship between Emotional Exhaustion and the Experienced Safety subscale of the EssenCES is fully mediated by fear of assault. These results suggest that changing the way staff perceive challenging behavior should reduce stress levels even though the objective experience of challenging behavior has not changed. In this case if reported fear of assault can be reduced then staff will experience less emotional exhaustion as a result of the same reported experience of challenging behavior. The experienced safety subscale of the EssenCES relates to the degree of perceived conflict and the threat of violence and aggression within the organization highly effective treatment cannot be imagined in an atmosphere of constant aggressive tension and threat of violence. This suggests that clear instructions about how to manage challenging events and increasing the support of staff may reduce emotional exhaustion (Rose, Harris, & Burns, 2010). It also suggests that training in relation to challenging behavior might change staff’s perception of events and reduce emotional exhaustion (e.g. Rose, 2010). While it is not suggested that this is a comprehensive model to understand the development of burnout and emotional exhaustion, the elements in the model do suggest a number of practical responses to helping staff manage their working lives better. It also suggests that organizations can develop programs of support for staff that recognize while some challenging behaviors are difficult to change interventions can have an effect by changing both environmental and organizational and staff cognitive variables (Phillips & Rose, 2010. Interventions in the future should take into account these variables routinely as part of a comprehensive assessment strategy. A number of issues need to be considered when trying to interpret the results of this research. For example, the EssenCES was a questionnaire that was developed for forensic settings and may not be applicable to community services for people with intellectual disabilities. However, the items on the questionnaire had relevance across both settings as it focused on issues such as relationships, emotions and the organization of services further adaptation or the development of a questionnaire specifically for community intellectual disability services may be more appropriate. Other difficulties with this research include the relatively low response rate; this may have been due to the fact that the research was targeting people who were specifically working with people who had challenging behavior. The survey was also relatively long, as a result many individuals may have thought that it was an extra burden to them. However, obtaining results that confirm mediation with a comparatively small sample do suggest that these are important relationships. The survey was also cross sectional and as a result it did not allow an exploration of testing the directionality of the relationships. Within the model, Mash and Johnston (1990) specifically suggest that the relationships are bidirectional. Some of the concepts within the model were similar for example, fear of assault and experienced safety, however they were measuring concepts from different perspectives that suggest staff are able to distinguish between how these issues relate to themselves and the broader organization. We did not control for demographic variables as these were not seen as central to the model however future research could investigate the impact of individual characteristics on these relationships.
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
947
Addditionally, as this study was a single-source design and work characteristics and well-being were measured at the same time, the influence of current mental state or situational influences may bias the evaluation of perceived stresses and/or strains. The relationship between these variables may thus be overestimated. References Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Dagnan, D., Trower, P., & Smith, R. (1998). Care staff responses to people with learning disabilities and challenging behaviour: A cognitive-emotional analysis. British Journal of Clinical Psychology, 37, 59–68. Devereux, J., Hastings, R., Noone, S., Firth, A., & Totsika, V. (2009). Social support and coping as mediators or moderators of the impact of work stressors on burnout in intellectual disability support staff. Research in Developmental Disabilities, 30, 367–377. Harris, P., Humphreys, J., & Thomson, G. (1994). A checklist of challenging behaviour: The development of a survey instrument. Mental Handicap Research, 7(2), 118–133. Hassall, R., Rose, J., & McDonald, J. (2005). Parenting stress in mothers of children with an intellectual disability: The effects of parental cognitions in relation to child characteristics and family support. Journal of Intellectual Disability Research, 49(6), 405–418. Hastings, R. P., Horne, S., & Mitchell, G. (2004). Burnout in direct care staff in intellectual disability services: A factor analytic study of the Maslach Burnout Inventory. Journal of Intellectual Disability Research, 48, 268–273. Hatton, C., Brown, R., Caine, A., & Emerson, E. (1995). Stressors, coping strategies and stress-related outcomes among direct care staff in staffed houses for people with learning disabilities. Mental Handicap Research, 8, 252–271. Hatton, C., Rose, J., & Rose, D. (2004). Researching staff. In E. Emerson, C. Hatton, T. Parmenter, & T. Thompson (Eds.), Handbook of methods for research and evaluation in intellectual disabilities. Chichester, England: Willey. Hatton, C., Wigham, S., & Craig, J. (2009). Developing measures of job performance for support staff in housing services for people with intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 22, 54–64. Hill, C., & Rose, J. (2009). Parenting stress in mothers of adults with intellectual disability: Parental cognitions in relation to child characteristics and family support. Journal of Intellectual Disability Research, 53(12), 696–980. Hill, C., & Rose, J. (2010). Parenting stress models and their application to parents of adults with intellectual disabilities. The British Journal of Developmental Disabilities, 56(110), 19–37. Howard, R., Rose, J., & Levinson, V. (2009). The psychological impact of violence on staff working with adults with intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 22(6), 538–548. Howells, K., Tonkin, M., Milburn, C., Lewis, J., Draycot, S., Cordwell, J., et al. (2009). The EssenCES measure of social climate: A preliminary validation and normative data in UK high secure hospital settings. Criminal Behaviour and Mental Health, 19, 308–320. Jenkins, R., Rose, J., & Lovell, C. (1998). Psychological well-being of staff working with people who have challenging behavior. Journal of Intellectual Disability Research, 41, 502–511. Jose, P. (2008). Workshop on statistical moderation and mediation. Presented at SASP Conference 2008. Leather, P., Beale, D., Lawrence, C., & Dickson, R. (1997). Effects of exposure to workplace violence and the mediating impact of fear. Work and Stress, 11(4), 329– 340. Mash, E. J., & Johnston, C. (1990). Determinants of parenting stress: Illustrations from families of hyperactive children and physically abused children. Journal of Clinical Child Psychology, 19, 313–328. Maslach, C., Jackson, S. E., & Leiter, M. P. (1996). Maslach Burnout Inventory Manual (3rd ed.). Palo Alto, CA: Consulting Psychologists Press, Inc. MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behavioral Research, 39(1), 99–128. Mills, S., & Rose, J. (2011). The relationship between challenging behavior, burnout and cognitive variables in staff working with people who have intellectual disabilities. Journal of Intellectual Disability Research, 55, 844–857. National Health Service. (2009). NHS health and well-being: Final Report. London: Department of Health. Available from: http://www.nhshealthandwellbeing.org/pdfs/NHS%20Staff%20H&WB%20Review%20Final%20Report%20VFinal%2020-11-09.pdf. Downloaded on the 26th September 2011. Phillips, N., & Rose, J. (2010). Predicting placement breakdown: Individual and environmental factors associated with the success or failure of community residential placements for adults with intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 23(3), 201–213. Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, and Computers, 36, 717–731. Rose, D., & Rose, J. (2005). Staff in services for people with intellectual disabilities: The impact of stress on attributions and challenging behaviour. Journal of Intellectual Disability Research, 49(11), 827–838. Rose, J. (1995). Stress and residential staff: An integration of existing research. Mental Handicap Research, 8(4), 220–236. Rose, J. (2010). Carer reports of the efficacy of cognitive behavioral interventions for anger. Research in Developmental Disabilities, 31, 1502–1508. Rose, J. (2011). How do staff psychological factors influence outcomes for people with developmental and intellectual disabilities in residential services? Current Opinion in Psychiatry, 24(5), 403–407. Rose, J., & Cleary, A. (2007). Care staff perceptions of challenging behaviour and fear of assault. Journal of Intellectual and Developmental Disabilities, 32(2), 153–161. Rose, J., David, G., & Jones, C. (2003). Staff who work with people who have intellectual disabilities: The importance of personality. Journal of Applied Research in Intellectual Disabilities, 16(4), 267–278. Rose, J., Jones, F., & Fletcher, B. C. (1998). Investigating the relationship between stress and worker behavior. Journal of Intellectual Disability Research, 42(2), 163– 172. Rose, J., Harris, P., & Burns, M. (2010). Supporting staff. In S. Hardy & J. Theresa (Eds.), Challenging behaviour: A training pack. Pavilion. Rose, J., Madurai, T., Thomas, K., Duffy, B., & Oyebode, J. (2010). Reciprocity and burnout in direct care staff. Clinical Psychology and Psychotherapy. Schalast, N., Redies, R., Collins, M., Stacey, J., & Howells, J. (2008). EssenCES, a short questionnaire for assessing the social climate of forensic psychiatric wards. Criminal Behavior and Mental Health, 18, 49–58. Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations models. In S. Leinhart (Ed.), Sociological methodology (pp. 290–312). San Francisco: Jossey–Bass. Thomas, K., & Rose, J. (2010). The relationship between reciprocity and the emotional and behavioural responses of staff. Journal of Applied Research in Intellectual Disabilities, 23(2), 167–179. Thompson, L., & Rose, J. (2011). Does organisational culture impact on burnout in staff who work with people with intellectual disabilities? A systematic review of the literature. Journal of Intellectual Disabilities, 15(3), 177–194. Weiner, B. (1985). An attributional theory of achievement, motivation and emotion. Psychological Review, 92, 547–573.
Contents lists available at SciVerse ScienceDirect
Research in Developmental Disabilities
Client characteristics, organizational variables and burnout in care staff: The mediating role of fear of assault John Rose *, Sophie Mills 1, Daniel Silva 2, Lauren Thompson 3 The University of Birmingham, Edgbaston, Birmingham, B15 2TT, United Kingdom
A R T I C L E I N F O
A B S T R A C T
Article history: Received 24 September 2012 Received in revised form 19 November 2012 Accepted 19 November 2012 Available online 3 January 2013
A broad range of factors have been identified as having an impact on burnout and performance. To improve our understanding of how these factors interact, a model of carer stress is tested. Staff were surveyed in residential units, assessments included burnout, organizational factors, staff cognitions and ratings of resident challenging behavior. The relationship between challenging behavior and emotional exhaustion was fully mediated by fear of assault. The relationship between emotional exhaustion and experienced safety (an organizational variable) was also fully mediated by fear of assault. The use of the model with staff is supported and it suggests that staff burnout can be reduced by influencing either staff cognitions, organizational factors or challenging behavior or a combination of these factors. ß 2012 Elsevier Ltd. All rights reserved.
Keywords: Burnout Challenging behavior Staff Cognitions Organizational factors Intellectual disability
1. Introduction Staff are increasingly being recognized as an important element in the care process (National Health Service, 2009). The psychological well-being of staff is seen as an important element in the delivery of care and support, and there is increasing evidence of a link between staff well-being and their effectiveness in providing support to people with intellectual disabilities (e.g. Hatton, Wigham, & Craig, 2009; Rose, Jones, & Fletcher, 1998). A range of variables have been identified as contributing to staff well being including client characteristics (e.g. Jenkins, Rose, & Lovell, 1998), organizational factors and staff cognitive variables (e.g. Rose, 2011). There is evidence to suggest that the way that staff perceive aspects of their working life, particularly challenging behavior can influence their reported levels of burnout. Howard, Rose, & Levinson (2009) have shown the relationship between challenging behavior and burnout can be moderated by the perceived self efficacy of staff. In a study based on a subset of the data presented here Mills and Rose (2011) have also shown the same relationship can be mediated by perceived fear of assault. That is staff report lower levels of burnout if they feel more competent, supported and consider the possibility of them being assaulted as lower than other staff. A similar relationship was found by Rose, David, and Jones (2003) in that wishful thinking (a coping strategy) partially mediated the relationship between work demands (of which challenging behavior could be an integral part) and stress. This relationship was confirmed by Devereux, Noone, Firth, and Totsika (2009) who found that wishful thinking was partially mediated by the relationship between work demands & emotional exhaustion in care staff.
* Corresponding author at: School of Psychology, The University of Birmingham, Edgbaston, Birmingham, B15 2TT, United Kingdom. Tel.: +44 01214142640. E-mail address: [email protected] (J. Rose). 1 Now at Black Country Partnership Foundation (NHS) Trust. 2 Now at University of Southampton. 3 Now at Bolton NHS foundation Trust. 0891-4222/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ridd.2012.11.014
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
941
Other studies have suggested that organizational variables contribute significantly to stress and burnout (Hatton, Brown, Caine, & Emerson, 1995; Thompson & Rose, 2011), and recent research that evaluates the differences between a variety of inputs and outputs in relation to psychological well-being suggests that staff rank organizational factors as the single largest contribution to staff well-being (Rose, Madurai, Thomas, Duffy, & Oyebode, 2010; Thomas & Rose, 2010). A complex array of factors can contribute to reported well-being and this complexity has contributed to the lack of a comprehensive model to explain the development of staff stress. A number of frameworks and models have been developed including Rose (1995) who proposed a simple framework relating to organizational proximity, and Hatton, Rose, and Rose (2004) who listed a broad range of core constructs that might influence the development of well-being, suggesting that staff morale and well-being was one construct that, among many, may directly influence well being. Other models have examined specific aspects of the relationship between staff and the people they care for such as attributions and have used Weiner’s (1985) model however, the results from these studies have been mixed (Dagnan, Trower, & Smith, 1998; Rose & Rose, 2005). Other models have been developed for specific purposes such as examining the importance of staff characteristics (Rose et al., 2003) or identifying factors that may be responsible for placement breakdown (Phillips & Rose, 2010). These models tend to be specific in focus and lack a broader overview so while they can contribute to answering specific questions they lack a broader conceptual overview. Alternative models can be found that examine the development of stress in parents of children with intellectual disabilities (e.g. Hill & Rose, 2010). While some of these models are complex and have been difficult to test empirically others have been partially tested and the results suggest that they have some utility. For example, a model of stress in parent child interactions based on Mash and Johnston (1990) is shown in Fig. 1. Mash and Johnston’s (1990) model consists of four elements including child characteristics such as ability and challenging behavior, environmental characteristics such as family support, parent characteristics that might include a variety of psychological variables such as attributions or locus of control with parent child interactive stress as an outcome. They see these links as reciprocal and while they recognize that other variables may be influential they note that these variables are likely to be the most important influence on the development of parent–child interactive stress. This model has been demonstrated to have predictive utility in mothers of children with intellectual disability (Hassall, Rose, & McDonald, 2005; Hill & Rose, 2009). A review of this model suggests that it may have direct applicability to understanding the development of staff well-being and burnout in that elements of the model map closely on to the factors that have been identified as important in research on staff well-being. The characteristics of clients are clearly of direct relevance, particularly challenging behavior.
Fig. 1. A model of stress in parent child interactions based on Mash and Johnston (1990).
942
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
Environmental or organizational characteristics have also been identified as important variables for staff (Thompson & Rose, 2011). A variety of staff variables have also been identified as important mediators between challenging behavior and wellbeing such as attributions, self efficacy and fear of assault. If effective measures of these elements of the model can be taken then their relationships can be investigated in relation to a modified version of the Mash and Johnston (1990) model. There are a range of commonly used measures of challenging behavior for people with Intellectual Disability (e.g. Checklist of Challenging Behaviors (Harris, Humphreys, & Thomson, 1994). A number of staff cognitive variables have also been shown to mediate the relationship between stress and challenging behavior including fear of assault, this is a concept developed by Leather, Beale, Lawrence, and Dickson (1997) who investigated levels of violence and fear of violence experienced by publicans in relation to their well-being. Measures of organizational characteristics are also being developed including the Essen Climate Evaluation Schema (EssenCES) which is a measure designed to assess organizational characteristics such as the social climate in forensic settings (Schalast, Redies, Collins, Stacey, & Howells, 2008) this assesses dimensions of social climate, including three subscales: ‘therapeutic hold’, ‘patient cohesion and mutual support’ and ‘experienced safety’. While these variables may not comprehensively represent all of the concepts originally considered by Mash and Johnston, they do provide a starting point for the empirical testing of the model. 1.1. Aim The aim of this paper is to use the structure provided by the Mash and Johnston (1990) model with direct care staff in residential settings. The model will be used to examine the relationship between staff psychological well being, challenging behavior and the therapeutic environment and how the relationship between these variables is mediated by a psychological variable, fear of assault. By doing this it is hoped that it will provide an initial test of the Mash and Johnston (1990) model with a view to gaining a greater understanding of how these complex factors interact. 2. Method 2.1. Design and sample The study was reviewed and given a favorable ethical opinion by a National Health Service (NHS) Local Research Ethics Committee. A cross-sectional correlational design was used. The participants included qualified and unqualified direct care staff who worked in residential homes for adults with intellectual disabilities. All of the staff were involved directly in the care of people with intellectual disabilities and would have spent the majority of their work time working with clients. Staff were required to have worked in their current place of work for at least three months. Staff working in a variety of services including both the National Health Service (NHS) and private sector were recruited, 78 staff took part from a total of six organizations however one was removed from the analysis as they had been at work for less than three months. The majority of staff worked in community settings, 14% of staff worked in secure settings. 2.2. Procedure Clinical psychologists were asked to identify homes where service users displayed challenging behavior and managers were approached to invite staff to take part in the study. While the focus of the study was on challenging behavior, in practice there was a very broad range of environments sampled from locations where considerable challenging behavior was experienced to other environments where residents rarely exhibited any challenging behavior. Questionnaire packs were then left with managers to distribute to the staff. Packs included a Participant Information Sheet, the Consent Form, and questionnaire. The questionnaire pack took approximately 30 min to complete. Participants either posted the questionnaires back to the researchers using a pre-paid envelope, or they were collected from the homes. 333 questionnaires were distributed and 78 were returned, a response rate of 23%. 2.3. Measures Fear of assault. The fear of assault measure comprised of two questions to measure staff’s fear of assault rated on a 5-point likert scale, adapted from Leather et al. (1997) by Rose and Cleary (2007). A Checklist of Challenging Behaviors. The Checklist of Challenging Behaviors (Harris et al., 1994), measures challenging behavior exhibited by a service user in the last three months. The participant is asked to rate on 5 point scales the frequency, management difficulty, and severity of 14 ‘aggressive behaviors’ (e.g. scratching, biting, kicking people) and 18 ‘other challenging behaviors’ (e.g. smashing windows, spitting, eating inappropriate things). For these other challenging behaviors the participant is only asked to rate the frequency and management difficulty, not severity. The checklist is a reliable measure, with significant critical values for rs at p < 0.05 using Spearman’s correlation coefficient for frequency, management difficulty, and severity. Maslach Burnout Inventory – Human Services Survey (MBI-HSS). (Maslach, Jackson, & Leiter, 1996) measures burnout in staff. Respondents complete statements rated from 0 (never) to 6 (every day) regarding their feelings about their
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
943
employment. The measure is divided into three scales; emotional exhaustion (EE), depersonalization (DP), and personal accomplishment (PA). A high level of burnout is characterized by high EE and DP scores, and low PA scores. Hastings, Horne, and Mitchell (2004) carried out a factor analytic study with staff working with people with intellectual disabilities the integrity of the scale and reliability were found to be acceptable (alpha: EE = 0.87, DP = 0.68, PA = 0.76). Modified version of Essen Climate Evaluation Schema. (EssenCES; Schalast et al., 2008). The EssenCES is a 15-item questionnaire that also uses a 5 point scale and provides a score related to the social climate in 3 domains: ‘therapeutic hold’, ‘patients’ cohesion and mutual support’, and ‘experienced safety.’ High scores in each of the domains represent a positive social climate. The EssenCES was designed to assess the social climate of forensic psychiatric wards, but it has been suggested that it could also be used in similar settings, including environments for people with challenging behavior. The questionnaire was adapted for example; the word ‘‘resident’’ was used instead of ‘‘patient’’. The unmodified EssenCES has reasonable internal consistency (therapeutic hold, a = 0.74; patient cohesion and mutual support, a = 0.78; experienced safety, a = 0.77; Howells et al., 2009). 2.4. Reliability To assure the internal consistency of the questionnaires and the ecological validity of this analysis, reliability was checked for two measures: EssenCES and the fear of assault. The Cronbach’s alpha for the three components of the EssenCES questionnaires were: therapeutic hold (0.73), experienced safety (0.80) and patient cohesion (0.80). In order to account for the number of items in the fear of assault questionnaire an inter item reliability was calculated that showed a positive correlation of r = 0.89. 3. Results The final sample consisted of 77 participants. There were also some missing data. See Tables 1 and 2 for demographic data from the sample. Means and standard deviations are reported for the main study variables in Table 3. Shapiro–Wilk tests were conducted and two of the scales failed the normality assumption: Patient Cohesion from EssenCES and other challenging behavior from the Checklist of Challenging Behavior. Table 1 Staff demographics: age and length of time in current and previous employment in intellectual disability services.
Age (years) Length of time in current employment (months) Length of time working in services for people with intellectual disabilities (months)
Mean
Range
SD
N
37 62 101
18–62 3–279 3–387
11.89 75.32 97.89
76 76 76
Table 2 Demographic data: gender, job title and training.
Gender Job title
Training related to challenging behavior
Male Female Support workers Qualified nursing staff Managerial positions (e.g. house/team leader) Other disciplines (including psychology, occupational therapy, social work) Breakaway/de-escalation Training relevant to challenging behavior (unspecified)
Percentage
N
29.87 70.13 63.16 11.84 14.47 10.53 19.05 4.76
77 76
63
Table 3 Means and standard deviations of scales.
Fear of assault Emotional exhaustion Depersonalisation Personal accomplishment Aggressive behavior Other behavior Patient cohesion Experienced safety Therapeutic hold
Minimum
Maximum
Mean
Std. deviation
2.00 0.00 0.00 19.00 1.00 1.00 0.00 0.00 7.00
10.00 51.00 26.00 48.00 3.98 4.22 19.00 18.00 20.00
5.3117 19.8649 4.7203 36.3959 2.1139 2.1611 8.6554 8.0682 15.7829
2.01481 11.09155 5.37931 6.67392 0.80512 0.70115 4.58028 4.75028 3.00343
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
944 Table 4 Correlation matrix for study variables. (1)
Measure Fear of assault (1) Emotional exhaustion (2) Depersonalization (3) Personal accomplishment (4) Aggressive behavior (5) Other behavior (6) Patient cohesion (7) Experienced safety (8) Therapeutic hold (9)
0.392
(2) **
(3) **
0.465 0.337**
(4) **
0.275 0.323** 0.206
(5) **
0.560 0.364** 0.295** 0.384**
(6) **
0.600 0.329** 0.410** 0.184 0.688**
(7) *
0.244 0.339** 0.219 0.343** 0.350** 0.369**
(8) **
0.451 0.266* 0.218 0.213 0.412** 0.439** 0.013
0.187 0.095 0.185 0.352** 0.268* 0.247* 0.378** 0.043
* Significant at P < 0.05. ** Significant at P < 0.01.
The analysis was performed in three steps. The first step considered the univariate relationships between the variables. Where the variables were normally distributed Pearson’s correlation was used and where the variables were not normally distributed Spearman’s Rho was calculated (Table 4). Second, an attempt was made to construct a pathway linking burnout, challenging behavior, environment measures and cognitive variable. Third, where the relationships indicated possible mediation, these processes were investigated. 3.1. Path analysis In order to construct a model of burnout considering all of the variables, the correlations between the measures for each variable were analyzed. The goal was to find pathways of significant correlations that linked measures of challenging behavior and environmental variables with the cognitive variable and burnout and that also linked the cognitive variable to burnout (see Fig. 2). A model would be formed if one of the measures from each of the variables followed these specifications. The data produced five models that fulfilled these conditions. The models are shown on Fig. 2 below.
Fig. 2. Adapted model of stress for staff–resident interactions showing only significant paths.
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
945
Table 5 Components of the models formed with the data by measure. Model
Challenging behavior
Organizational characteristics
Cognitive variables
Burnout
1 2 3 4 5
Aggressive Aggressive Aggressive Other Other
Patient cohesion Patient cohesion Experienced safety Patient cohesion Experienced safety
Fear Fear Fear Fear Fear
Emotional exhaustion Personal accomplishment Emotional exhaustion Emotional exhaustion Emotional exhaustion
of of of of of
assault assault assault assault assault
3.2. Mediation Although the models indicate a relationship between the variables, a formal analysis for mediation is necessary to accurately understand their scope. Mediation analyses are often guided by the assumptions proposed by Baron and Kenny (1986): the independent variable (IV) significantly affects the mediator, the IV significantly affects the dependent variable (DV) in the absence of the mediator, the mediator has a significant unique effect on the DV and finally, the effect of the IV on the DV reduces upon the addition of the mediator to the model. However, Baron & Kenny’s model for mediation is quite conservative in defining full mediation as when the effect of the IV on the DV in the presence of the mediator diminishes to zero. This scenario is unrealistic for the social sciences (Jose, 2008). Thus, it is recommended that mediational analyses be based on formal significance tests of the indirect effect. The most used formal analysis of mediation is the Sobel test (Sobel, 1982), this approach is more powerful than the stepwise procedure proposed by Baron and Kenny because it addresses mediation directly. Although useful, for this particular analysis the Sobel test cannot be used because the variables are not normally distributed. Therefore, bootstrapping is recommended. In such procedure, random amounts of indirect effects are extracted from the data and the confidence interval for each of these samples is calculated to inform what would be the confidence interval for the whole data. Through the application of bootstrapped confidence intervals, it is possible to avoid power problems introduced by asymmetric sampling distributions of an indirect effect (MacKinnon, Lockwood, & Williams, 2004). Preacher and Hayes (2004) developed a macro script to formally evaluate full mediation through bootstrapping. The matrix gives you the result for the four assumptions from Baron and Kenny as well as the confidence interval for both 95% and 99% confidence. With such analysis to reject the null hypothesis (assume significant mediation) zero cannot be within your confidence interval. This would mean that there is a possibility that the mediation effect could be due to chance. In order for the macro script to be used, the models were split in two subsets. On one side, the direct and indirect effect of challenging behavior on burnout and on the other side the direct and indirect effect of environment on burnout. This was Table 6 Summary of regression analysis between experienced safety and emotional exhaustion mediated by fear of assault. Regression
Independent variables
Dependent variables
1 2 3
Experienced safety Fear of assault Experienced safety Fear of assault Emotional exhaustion experienced safety
Emotional exhaustion Emotional exhaustion Emotional exhaustion
0.12 (0.05) 0.08 (0.02) 0.99 (0.30)
2.61 4.33 3.40
0.01** 0.00** 0.00**
Fear of assault
0.049 (0.05)
0.94
0.35
4 Bootstrapping effect
Mean 0.072
SE 0.028
Beta (S.E.)
t
p
LL 99% CI
UL 99% CI
0.166
.013
Note. n = 77 participants. Unstandardized regression coefficients are reported. Bootstrap sample size = 10,000. LL, lower limit; CI, confidence interval; UL, upper limit.
Table 7 Summary of regression analysis between aggressive behavior and emotional exhaustion mediated by fear of assault. Regression
Independent variables
Dependent variables
Beta (S.E.)
t
p
1 2 3
Aggressive behavior Fear of assault Aggressive behavior Fear of assault Emotional exhaustion aggressive behavior
Emotional exhaustion Emotional exhaustion Emotional exhaustion
0.03 (.07) 0.08 (.02) 0.18 (.04)
4.09 4.33 4.23
0.00** 0.00** 0.00**
Fear of assault
0.02 (.00)
2.11
0.03**
4
Bootstrapping effect
M
SE
LL 99% CI
UL 99% CI
0.154
0.005
0.003
0.032
Note. n = 77 participants. Unstandardized regression coefficients are reported. Bootstrap sample size = 10,000. LL, lower limit; CI, confidence interval; UL, upper limit.
946
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
Table 8 Summary of regression analysis between other challenging behavior and emotional exhaustion mediated by fear of assault. Regression
Independent variables
Dependent variables
Beta (S.E.)
t
p
1 2 3
Other challenging behavior Fear of assault Other challenging behavior Fear of assault Emotional exhaustion other challenging behavior
Emotional exhaustion Emotional exhaustion Emotional exhaustion
0.02 (0.01) 0.08 (0.02) 0.19 (0.03)
3.55 4.33 5.29
0.00** 0.00** 0.00**
Fear of assault
0.01 (0.01)
1.23
0.22
4 Bootstrapping effect
M
SE
LL 99% CI
UL 99% CI
0.160
0.005
0.005
0.031
Note. n = 77 participants. Unstandardized regression coefficients are reported. Bootstrap sample size = 10,000. LL, lower limit; CI, confidence interval; UL, upper limit.
necessary because the program could only account for one dependent variable, one independent variable and one mediator at a time. However, for the interpretation of the results, a model would only be considered significant if both subsets showed significant results. The bootstrapping results showed no significant mediation between the following subsets: patient cohesion – fear of assault – emotional exhaustion (CI at 95% = 0.07, 0.02) and patient cohesion – fear of assault – personal accomplishment (CI at 95% = 0.01, 0.07). Due to these results, models one, two and four (Table 5) were disregarded. Tables 6–8 show the results for the three regressions suggested by Baron and Kenny and the bootstrapping analysis for the significant subsets. The significant indirect effect indicates full mediation for the following models: aggressive challenging behavior, experienced safety, fear of assault and emotional exhaustion; and other challenging behavior, experienced safety, fear of assault and emotional exhaustion. 4. Discussion The results suggest that Mash and Johnston’s (1990) model may have utility as burnout can be understood as a corollary of challenging behavior, organizational climate and cognitive variables. The influence of challenging behavior including both aggression and other challenging behavior, as measured by the Checklist of Challenging Behavior on emotional exhaustion is fully mediated by fear of assault. However, only the relationship between Emotional Exhaustion and the Experienced Safety subscale of the EssenCES is fully mediated by fear of assault. These results suggest that changing the way staff perceive challenging behavior should reduce stress levels even though the objective experience of challenging behavior has not changed. In this case if reported fear of assault can be reduced then staff will experience less emotional exhaustion as a result of the same reported experience of challenging behavior. The experienced safety subscale of the EssenCES relates to the degree of perceived conflict and the threat of violence and aggression within the organization highly effective treatment cannot be imagined in an atmosphere of constant aggressive tension and threat of violence. This suggests that clear instructions about how to manage challenging events and increasing the support of staff may reduce emotional exhaustion (Rose, Harris, & Burns, 2010). It also suggests that training in relation to challenging behavior might change staff’s perception of events and reduce emotional exhaustion (e.g. Rose, 2010). While it is not suggested that this is a comprehensive model to understand the development of burnout and emotional exhaustion, the elements in the model do suggest a number of practical responses to helping staff manage their working lives better. It also suggests that organizations can develop programs of support for staff that recognize while some challenging behaviors are difficult to change interventions can have an effect by changing both environmental and organizational and staff cognitive variables (Phillips & Rose, 2010. Interventions in the future should take into account these variables routinely as part of a comprehensive assessment strategy. A number of issues need to be considered when trying to interpret the results of this research. For example, the EssenCES was a questionnaire that was developed for forensic settings and may not be applicable to community services for people with intellectual disabilities. However, the items on the questionnaire had relevance across both settings as it focused on issues such as relationships, emotions and the organization of services further adaptation or the development of a questionnaire specifically for community intellectual disability services may be more appropriate. Other difficulties with this research include the relatively low response rate; this may have been due to the fact that the research was targeting people who were specifically working with people who had challenging behavior. The survey was also relatively long, as a result many individuals may have thought that it was an extra burden to them. However, obtaining results that confirm mediation with a comparatively small sample do suggest that these are important relationships. The survey was also cross sectional and as a result it did not allow an exploration of testing the directionality of the relationships. Within the model, Mash and Johnston (1990) specifically suggest that the relationships are bidirectional. Some of the concepts within the model were similar for example, fear of assault and experienced safety, however they were measuring concepts from different perspectives that suggest staff are able to distinguish between how these issues relate to themselves and the broader organization. We did not control for demographic variables as these were not seen as central to the model however future research could investigate the impact of individual characteristics on these relationships.
J. Rose et al. / Research in Developmental Disabilities 34 (2013) 940–947
947
Addditionally, as this study was a single-source design and work characteristics and well-being were measured at the same time, the influence of current mental state or situational influences may bias the evaluation of perceived stresses and/or strains. The relationship between these variables may thus be overestimated. References Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Dagnan, D., Trower, P., & Smith, R. (1998). Care staff responses to people with learning disabilities and challenging behaviour: A cognitive-emotional analysis. British Journal of Clinical Psychology, 37, 59–68. Devereux, J., Hastings, R., Noone, S., Firth, A., & Totsika, V. (2009). Social support and coping as mediators or moderators of the impact of work stressors on burnout in intellectual disability support staff. Research in Developmental Disabilities, 30, 367–377. Harris, P., Humphreys, J., & Thomson, G. (1994). A checklist of challenging behaviour: The development of a survey instrument. Mental Handicap Research, 7(2), 118–133. Hassall, R., Rose, J., & McDonald, J. (2005). Parenting stress in mothers of children with an intellectual disability: The effects of parental cognitions in relation to child characteristics and family support. Journal of Intellectual Disability Research, 49(6), 405–418. Hastings, R. P., Horne, S., & Mitchell, G. (2004). Burnout in direct care staff in intellectual disability services: A factor analytic study of the Maslach Burnout Inventory. Journal of Intellectual Disability Research, 48, 268–273. Hatton, C., Brown, R., Caine, A., & Emerson, E. (1995). Stressors, coping strategies and stress-related outcomes among direct care staff in staffed houses for people with learning disabilities. Mental Handicap Research, 8, 252–271. Hatton, C., Rose, J., & Rose, D. (2004). Researching staff. In E. Emerson, C. Hatton, T. Parmenter, & T. Thompson (Eds.), Handbook of methods for research and evaluation in intellectual disabilities. Chichester, England: Willey. Hatton, C., Wigham, S., & Craig, J. (2009). Developing measures of job performance for support staff in housing services for people with intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 22, 54–64. Hill, C., & Rose, J. (2009). Parenting stress in mothers of adults with intellectual disability: Parental cognitions in relation to child characteristics and family support. Journal of Intellectual Disability Research, 53(12), 696–980. Hill, C., & Rose, J. (2010). Parenting stress models and their application to parents of adults with intellectual disabilities. The British Journal of Developmental Disabilities, 56(110), 19–37. Howard, R., Rose, J., & Levinson, V. (2009). The psychological impact of violence on staff working with adults with intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 22(6), 538–548. Howells, K., Tonkin, M., Milburn, C., Lewis, J., Draycot, S., Cordwell, J., et al. (2009). The EssenCES measure of social climate: A preliminary validation and normative data in UK high secure hospital settings. Criminal Behaviour and Mental Health, 19, 308–320. Jenkins, R., Rose, J., & Lovell, C. (1998). Psychological well-being of staff working with people who have challenging behavior. Journal of Intellectual Disability Research, 41, 502–511. Jose, P. (2008). Workshop on statistical moderation and mediation. Presented at SASP Conference 2008. Leather, P., Beale, D., Lawrence, C., & Dickson, R. (1997). Effects of exposure to workplace violence and the mediating impact of fear. Work and Stress, 11(4), 329– 340. Mash, E. J., & Johnston, C. (1990). Determinants of parenting stress: Illustrations from families of hyperactive children and physically abused children. Journal of Clinical Child Psychology, 19, 313–328. Maslach, C., Jackson, S. E., & Leiter, M. P. (1996). Maslach Burnout Inventory Manual (3rd ed.). Palo Alto, CA: Consulting Psychologists Press, Inc. MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behavioral Research, 39(1), 99–128. Mills, S., & Rose, J. (2011). The relationship between challenging behavior, burnout and cognitive variables in staff working with people who have intellectual disabilities. Journal of Intellectual Disability Research, 55, 844–857. National Health Service. (2009). NHS health and well-being: Final Report. London: Department of Health. Available from: http://www.nhshealthandwellbeing.org/pdfs/NHS%20Staff%20H&WB%20Review%20Final%20Report%20VFinal%2020-11-09.pdf. Downloaded on the 26th September 2011. Phillips, N., & Rose, J. (2010). Predicting placement breakdown: Individual and environmental factors associated with the success or failure of community residential placements for adults with intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 23(3), 201–213. Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, and Computers, 36, 717–731. Rose, D., & Rose, J. (2005). Staff in services for people with intellectual disabilities: The impact of stress on attributions and challenging behaviour. Journal of Intellectual Disability Research, 49(11), 827–838. Rose, J. (1995). Stress and residential staff: An integration of existing research. Mental Handicap Research, 8(4), 220–236. Rose, J. (2010). Carer reports of the efficacy of cognitive behavioral interventions for anger. Research in Developmental Disabilities, 31, 1502–1508. Rose, J. (2011). How do staff psychological factors influence outcomes for people with developmental and intellectual disabilities in residential services? Current Opinion in Psychiatry, 24(5), 403–407. Rose, J., & Cleary, A. (2007). Care staff perceptions of challenging behaviour and fear of assault. Journal of Intellectual and Developmental Disabilities, 32(2), 153–161. Rose, J., David, G., & Jones, C. (2003). Staff who work with people who have intellectual disabilities: The importance of personality. Journal of Applied Research in Intellectual Disabilities, 16(4), 267–278. Rose, J., Jones, F., & Fletcher, B. C. (1998). Investigating the relationship between stress and worker behavior. Journal of Intellectual Disability Research, 42(2), 163– 172. Rose, J., Harris, P., & Burns, M. (2010). Supporting staff. In S. Hardy & J. Theresa (Eds.), Challenging behaviour: A training pack. Pavilion. Rose, J., Madurai, T., Thomas, K., Duffy, B., & Oyebode, J. (2010). Reciprocity and burnout in direct care staff. Clinical Psychology and Psychotherapy. Schalast, N., Redies, R., Collins, M., Stacey, J., & Howells, J. (2008). EssenCES, a short questionnaire for assessing the social climate of forensic psychiatric wards. Criminal Behavior and Mental Health, 18, 49–58. Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations models. In S. Leinhart (Ed.), Sociological methodology (pp. 290–312). San Francisco: Jossey–Bass. Thomas, K., & Rose, J. (2010). The relationship between reciprocity and the emotional and behavioural responses of staff. Journal of Applied Research in Intellectual Disabilities, 23(2), 167–179. Thompson, L., & Rose, J. (2011). Does organisational culture impact on burnout in staff who work with people with intellectual disabilities? A systematic review of the literature. Journal of Intellectual Disabilities, 15(3), 177–194. Weiner, B. (1985). An attributional theory of achievement, motivation and emotion. Psychological Review, 92, 547–573.